[
https://issues.apache.org/jira/browse/METRON-694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15892518#comment-15892518
]
ASF GitHub Bot commented on METRON-694:
---------------------------------------
Github user JonZeolla commented on the issue:
https://github.com/apache/incubator-metron/pull/453
I agree, I [recommended the same
thing](https://lists.apache.org/thread.html/2a673bc97d975bc7e8e160228742c07a8cf45e43c1b1efb3fea83579@%3Cdev.metron.apache.org%3E)
on the dev list and in IRC in the past. In the case where the TTL expires, we
should do *something*. Maybe we attempt to place a standard, simple message to
be indexed which has a very high likelihood of successfully being stored and
that is exempt from any sort of retries or TTLs.
That said, there are no other specific looping scenarios that I'm currently
able to envision, which is why I wasn't pushing that concept further for this
PR. Given the one that we are aware of, if we only took the TTL route, it
would just loop and then fail the entire message, where adding `"ignore_above":
8191` would never loop and only removes the bare minimum from the message in
order to get it to properly index, for the given known problem scenario.
> Index Errors from Topologies
> ----------------------------
>
> Key: METRON-694
> URL: https://issues.apache.org/jira/browse/METRON-694
> Project: Metron
> Issue Type: Bug
> Reporter: Ryan Merriman
>
> Need to make sure (and review) that all the bolts write into the error queue.
> Errors should then be consumed from the error queue and indexed.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
